The Power of AI to Revolutionize the World, But It May Not Always Pay Off

What if artificial intelligence (AI) revolutionizes nearly everything, yet still turns out to be one of the worst financial decisions ever made?

This may seem paradoxical, but it isn’t. Technology can reshape society while the firms that develop it overspend, leaving investors to cover the losses.

Listen/read:
Tech bubbles, AI, and emerging markets
Investing in a world gone tech-crazy

The internet exemplified this phenomenon: society benefited even as many investors faced losses.

This scenario looms over AI. Top tech companies in the United States (US) could invest $900 billion in chips, data centers, and energy in 2026, increasing to $1.4 trillion in 2027. However, AI revenue projections are only between $150 billion and $220 billion, suggesting the industry might require $2.5 trillion annually to sustain its spending.

This does not imply that AI is a scam or a bubble. Similar to railways, electricity, and the internet, substantial investments were essential before the benefits of productivity became evident.

Nevertheless, transformative infrastructures often offer greater rewards to society than to their financiers. Technology can indeed change everything while overspending, intense competition, and falling prices undermine the financial viability for those funding it.

China highlights this tension vividly: Its companies are investing less than one-tenth of what their US counterparts allocate for data centers, yet their superior models are nearing the frontier. Factors like lower land and labor costs aid in this achievement, along with advancements in distillation and engineering.

Chinese squeeze

Interestingly, US-imposed chip restrictions have compelled Chinese developers to extract more performance from limited computing resources.

This scarcity has enforced discipline, but it has also led to waiting lists, rationed services, and extended processing times. China demonstrates that frugality does not always conquer, nor is spending irrelevant. The vital measure of AI’s success lies not in how intelligent a model seems, but in the useful output generated per dollar invested.

The US might not be passing this evaluation. Companies are acquiring AI technologies without first restructuring themselves to accommodate it. Ninety percent of executives report no increase in productivity over the past three years.

What’s lacking isn’t another chatbot; it’s clean data, revamped workflows, training for employees, and management willing to abandon outdated processes. Otherwise, AI merely enables workers to carry out yesterday’s tasks more efficiently.

Read/listen:
Meta’s Zuckerberg defends AI investments to skeptical investors
AI Wave: CapEx could change market dynamics
Alphabet’s $205bn spending target raises AI cost concerns

India illustrates the implications for workers. While its technology workforce has not diminished, routine outsourced jobs are declining, and in-house capability centers are expanding. The demand is shifting from general IT skills to individuals who blend AI expertise, technical skills, and business acumen.

The concern is not that all jobs will vanish, but that entry-level roles may disappear before younger workers accumulate the experience for more valuable positions.

Then, there’s South Korea, where AI demand has spurred chip profits, bonuses, and tax revenues. Its stock market surged, followed by a significant correction as concentration and borrowed funds amplified the downturn. Once again, a remarkable industrial story transformed into a risky venture. Sellers of tools may gain immense wealth during a gold rush, yet their investors can still pay a premium for that wealth.

Read:
Scarred by ‘97 Crash, Korea cautiously opens 24-hour won trading
Investors seeking refuge from the AI storm are turning to India
Korea overtakes the UK to become the world’s eighth-largest stock market

This holds significance for South Africans, whose global investments are increasingly tied to US tech stocks. Investors must discern three questions: Is AI genuinely transformative? Which companies will harness its value? Is that value already integrated into the pricing? A simple “yes” to the first question provides little insight into the latter two.

Where SA can compete … 

South Africa cannot outpace the US in constructing data centers, surpass China in capital investments, or out-manufacture Korea in chip production.

Ironically, this scarcity of resources may benefit us by compelling a focus on application: Leveraging AI in financial services, mining, agriculture, logistics, and healthcare; reducing service costs for underserved clients; and increasing productivity per employee.

In a nation facing substantial unemployment, success cannot equate to automating as many jobs as possible. It must involve enhancing output, crafting new markets, and empowering workers to gain greater value.

Read/listen:
eThekwini reports the lowest official unemployment rate of any South African metro
SA’s tax revolt is already underway, says economist
Unemployment rate rises to a four-year high of 33.6%

Each AI investment should address three practical inquiries: What costly issue will be eliminated? What new revenue streams are feasible? When will the return exceed the total cost of restructuring the organization?

The AI race won’t be determined by who spends the most but by who wastes the least.

* Dr Francois Stofberg is a financial well-being economist at the Efficient Group.